Three hikers rescued from Mount Shasta after Google Gemini told them to bring too little food and water

Key points

• Three hikers needed rescue from California's Mount Shasta after relying on Google Gemini, an AI chatbot, for trip planning.
• The Siskiyou County sheriff's office said Gemini advised the group to bring far less food and water than they needed.
• The hikers reached the summit at 7pm, seven hours past the recommended turnaround time of noon.
• Rangers and volunteers rescued the group the following morning after they spent the night in Mud Creek Canyon.

Three young men set out to climb California's Mount Shasta at 3am, armed with advice from Google Gemini, an AI chatbot that can answer questions and give recommendations through conversation. The trip went badly wrong. They did not make the summit until 7pm, a full seven hours after the noon cutoff that guides and rangers tell hikers to treat as a hard turnaround point.

With darkness closing in, the trio tried to descend anyway. They called the Siskiyou County sheriff's office for directions. They spent the night in Mud Creek Canyon. The next morning, Forest Service rangers and volunteers brought them out.

What did the AI actually get wrong?

The sheriff's office was specific. Gemini advised the hikers to bring far less food and water than their group needed, a shortfall that turned dangerous once their planned eight-hour climb stretched into a two-day ordeal.

It is worth being clear about what an AI chatbot can and cannot do here. Gemini, like other large language models (the technology behind tools such as ChatGPT, software that predicts useful text based on vast amounts of training data), draws on general information it has seen. It does not know current trail conditions, local weather forecasts, or the specific fitness level of the people asking. A ranger does.

"It is always advisable to call the local USFS Mount Shasta ranger station ahead of your trip," the sheriff's office said in its statement, "to ensure you have the most accurate information, and to never rely solely on AI for your trip planning."

Should you still use AI to plan outdoor trips?

AI tools can genuinely help with logistics: researching gear lists, understanding trail difficulty ratings, or drafting a packing checklist. That is real value. But local, real-time knowledge is where they fall short, and in wilderness settings that gap can hurt you.

The sheriff's story is also worth reading carefully before you file it under "stupid hikers." Survivorship bias runs through most AI cautionary tales. For every rescue that makes the news, thousands of people use chatbots for trip research without incident. The lesson is not to avoid AI entirely. It is to treat it like a well-read friend who has never actually climbed this particular mountain.

First reported by TechCrunch AI, the incident has renewed calls from search-and-rescue professionals for hikers to verify AI-sourced advice with local experts before they leave the trailhead.

Common questions

Can AI chatbots give bad advice that puts people in danger?

Yes, particularly when the question requires local, current, or highly specific knowledge. AI chatbots work from general training data and cannot check real-time conditions, so their answers can be plausible-sounding but wrong in ways that matter.

What is the safest way to use AI for outdoor trip planning?

Use it for general research and initial checklists, then verify every critical detail (water needs, turnaround times, current trail status) with the relevant ranger station or local guide before you go.

Who should I actually call before a big hike?

The land management agency responsible for your trail. For Mount Shasta, that is the US Forest Service Mount Shasta ranger station. A five-minute call can surface conditions no chatbot has access to.

Honest takeaway: AI is a useful starting point for trip research. It is not a substitute for a phone call to the ranger station.